clawd/tmp/contacts/postprocess.py

#!/usr/bin/env python3
"""
Post-process merged.vcf:
  1. Remove contacts with no email AND no phone
  2. Remove contacts with no name (FN or ORG) — list their emails for review
  3. Convert Dutch phone numbers (0X...) to international format (+31...)
"""

import re, sys

INPUT  = "/home/johan/clawd/tmp/contacts/merged.vcf"
OUTPUT = "/home/johan/clawd/tmp/contacts/final.vcf"

# ── vCard parser (simple block-based) ─────────────────────────────────────────

def parse_blocks(path):
    with open(path, encoding='utf-8', errors='replace') as f:
        raw = re.sub(r'\r?\n[ \t]', '', f.read())
    blocks = []
    for block in re.split(r'(?=BEGIN:VCARD)', raw, flags=re.I):
        block = block.strip()
        if block.upper().startswith('BEGIN:VCARD'):
            blocks.append(block)
    return blocks

def block_lines(block):
    return [l for l in block.splitlines() if l.strip() and l.upper() not in ('BEGIN:VCARD', 'END:VCARD', 'VERSION:3.0')]

def get_field_val(block, prefix):
    """First value matching field prefix."""
    for line in block.splitlines():
        k, _, v = line.partition(':')
        if re.match(r'(ITEM\d+\.)?' + prefix, k.strip(), re.I):
            return v.strip()
    return ''

def has_email(block):
    return bool(re.search(r'^(ITEM\d+\.)?EMAIL\b', block, re.I | re.M))

def has_phone(block):
    return bool(re.search(r'^(ITEM\d+\.)?TEL\b', block, re.I | re.M))

def get_name(block):
    fn = get_field_val(block, 'FN')
    if fn:
        return fn
    org = get_field_val(block, 'ORG')
    return org

def get_emails(block):
    emails = []
    for line in block.splitlines():
        k, _, v = line.partition(':')
        if re.match(r'(ITEM\d+\.)?EMAIL', k.strip(), re.I) and v.strip():
            emails.append(v.strip().lower())
    return emails

# ── Dutch phone normaliser ─────────────────────────────────────────────────────

def nl_to_intl(phone):
    """
    Convert Dutch local format to E.164-ish international format.
    Rules:
      - Already international (+XX or 00XX) → leave alone
      - US numbers (10 digit, starts with 1 after stripping) → leave alone
      - 10-digit Dutch numbers starting with 0 → replace leading 0 with +31
      Handles spaces/dashes/dots in input, preserves spacing style.
    """
    p = phone.strip()

    # Already international
    if p.startswith('+') or p.startswith('00'):
        return p

    # Strip formatting to count digits
    digits_only = re.sub(r'\D', '', p)

    # Must start with 0 and be 10 digits to be Dutch local
    if not digits_only.startswith('0') or len(digits_only) != 10:
        return p

    # Replace leading 0 with +31, preserve the rest of the formatting
    # e.g. "06 23 123 456" → "+31 6 23 123 456"
    # e.g. "0646438755"    → "+31646438755"
    # e.g. "020-1234567"   → "+3120-1234567"  (Amsterdam landline)
    converted = '+31' + p[1:]  # drop leading '0', prepend +31
    return converted

def convert_phones_in_block(block):
    lines = []
    for line in block.splitlines():
        k, _, v = line.partition(':')
        if re.match(r'(ITEM\d+\.)?TEL', k.strip(), re.I) and v.strip():
            new_v = nl_to_intl(v.strip())
            if new_v != v.strip():
                line = f'{k}:{new_v}'
        lines.append(line)
    return '\n'.join(lines)

# ── main ──────────────────────────────────────────────────────────────────────

def main():
    blocks = parse_blocks(INPUT)
    print(f"Input: {len(blocks)} contacts")

    kept = []
    removed_no_contact = 0
    removed_no_name_emails = []  # emails of nameless contacts

    phones_converted = 0

    for block in blocks:
        # Step 1: must have email or phone
        if not has_email(block) and not has_phone(block):
            removed_no_contact += 1
            continue

        # Step 2: must have a name
        name = get_name(block)
        if not name or not name.strip():
            emails = get_emails(block)
            removed_no_name_emails.extend(emails if emails else ['(no email — phone only)'])
            continue

        # Step 3: Dutch phone normalisation
        new_block = convert_phones_in_block(block)
        if new_block != block:
            phones_converted += 1
        block = new_block

        kept.append(block)

    print(f"Removed (no email+phone): {removed_no_contact}")
    print(f"Removed (no name):        {len(removed_no_name_emails)} email(s) from nameless contacts")
    print(f"Phone numbers converted:  {phones_converted}")
    print(f"Output:                   {len(kept)} contacts")

    # Write final vcf
    with open(OUTPUT, 'w', encoding='utf-8') as f:
        for block in kept:
            if not block.startswith('BEGIN:VCARD'):
                block = 'BEGIN:VCARD\nVERSION:3.0\n' + block
            if not block.endswith('END:VCARD'):
                block = block + '\nEND:VCARD'
            f.write(block + '\n\n')

    print(f"\nWritten to: {OUTPUT}")

    if removed_no_name_emails:
        print(f"\n── Nameless contacts (emails for review) ──────────────────")
        for e in sorted(set(removed_no_name_emails)):
            print(f"  {e}")

if __name__ == '__main__':
    main()